10 The Mantel–Haenszel procedure
Mantel and Haenszel (1959) proposed, for case-control studies of cancer stratified by age and other factors, a summary odds ratio and a test of association that need no iteration, tolerate empty cells and remain valid when the strata are many and small. Their methods, extended since to risk ratios, risk differences and rate ratios, are the standard tools of stratified analysis. Throughout, stratum \(k\) has the table of Section 4 with cells \(a_k, b_k, c_k, d_k\), margins \(n_{1k}, n_{0k}, m_{1k}, m_{0k}\) and total \(n_k\).
10.1 The Mantel–Haenszel odds ratio
10.2 Variance and confidence interval
10.3 The Mantel–Haenszel test
10.4 Matched pairs
10.5 Risk ratios and risk differences
10.6 Rate ratios
10.7 Testing homogeneity: the Breslow–Day test
10.8 Problems
10.2 Variance and confidence interval
10.3 The Mantel–Haenszel test
10.4 Matched pairs
10.5 Risk ratios and risk differences
10.6 Rate ratios
10.7 Testing homogeneity: the Breslow–Day test
10.8 Problems
Questions on this section
Stuck on something here? Ask below and it stays attached to this topic.